Valvular heart disease affects more than 100 million people worldwide and an estimated 300000 prosthetic heart valves are implanted every year. This rate is increasing, mainly driven by intrinsic valvular degeneration in aging populations.1

The use of bioprosthetic valve replacement (BVR) has also increased in the last 20 years compared to mechanical heart valves. This shift is difficult to explain, but may be related to the drawbacks associated with long-term oral anticoagulation (OAC) in younger patients as well as the higher burden of side effects, particularly in the elderly.2 Even with appropriate OAC therapy after valve implantation, there is a higher lifelong risk of thromboembolic events (estimated at 1–4%) and bleeding (2–9%) compared to the general population.3,4

Non-vitamin K antagonist oral anticoagulants (NOACs) have been shown to be safe and at least as effective as vitamin K antagonists (VKAs) in patients with atrial fibrillation (AF) and without moderate to severe mitral stenosis or rheumatic valvular disease (RVD).5 However, patients with mechanical heart valves should be anticoagulated with VKAs, given the harmful effects of dabigatran noted in the RE-ALIGN phase II trial, and more recently, the premature termination of the PROACT Xa trial of apixaban versus warfarin.5–7

However, the ideal OAC strategy in patients with mitral BVR or mitral valve repair (MVR) remains a matter of debate in the cardiovascular community, especially regarding the first three months after surgery. The American College of Cardiology/American Heart Association (ACC/AHA) guidelines recommend initiating OAC with VKAs within three months of index valvular surgery regardless of rhythm or valve position, and the European Society of Cardiology/European Association for Cardio-Thoracic Surgery (ESC/EACTS) guidelines also recommend VKAs for biological heart valves in mitral position within three months, adding that NOACs may be considered for patients with AF and a bioprosthesis in mitral position (class of recommendation IIb, level of evidence C). These recommendations are mainly based on observational studies.8–12

Nevertheless, off-label use of NOACs within three months of index mitral surgery is growing, supported by small studies and meta-analyses.4,13–15

In the RIVER trial, the subgroup of patients taking rivaroxaban in the first months after index valvular surgery showed fewer primary endpoint events, but only 18.8% of the patients in the main study were randomized within three months.16 Similarly, the ENAVLE trial showed that edoxaban was non-inferior to warfarin regarding the primary efficacy endpoint as well as for major bleeding. However, this was a small trial which included patients with bioprosthetic aortic valve replacement (49%), which may limit the applicability of these results.17

Regarding thromboembolic and bleeding events, NOACs appear to be as safe and as effective as VKAs in patients with surgical mitral BVR or MVR within three months of surgery. However, stronger evidence from observational studies and larger randomized trials are needed to further sustain these findings.

This study aimed to evaluate the efficacy and safety of NOACs in the first three months after MVR or mitral BVR compared to VKAs.

This study found that treatment with NOACs within three months of bioprosthetic mitral valve implantation or mitral repair was associated with fewer net clinical events, mainly driven by less bleeding. In addition, switching OAC strategy from NOACs to VKAs at discharge in AF patients was associated with more bleeding events. Therefore, these results suggest that NOACs were as safe and as effective as VKAs in this patient cohort.

In the present study, cardiovascular risk factors such as hypertension, dyslipidemia and diabetes resembled those of the groups studied in the RIVER and ENAVLE trials.16,17 From a surgical standpoint, 51.4% of patients underwent mitral BVR and 48.6% MVR; AF was previously diagnosed in 40.5% of patients.

At discharge, 98 (66.2%) patients were on VKAs and 50 (33.8%) on NOACs for at least three months. In the subgroup of AF patients with previous indication for OAC, therapy on admission influenced the OAC strategy on discharge, with 21 patients (55.3%) in the VKA group and 20 (95.2%) in the NOAC group discharged under the same therapy prescribed on admission. Switching OAC on discharge was infrequent (18 patients, 30%), albeit more common in the VKA group (43.6% patients switched from NOACs to VKAs on discharge, p=0.002).

The primary outcome occurred in 22 (22.4%) patients in the VKA group and in three (6%) in the NOAC group (p=0.012), mainly driven by more bleeding events in the former (19.4%, p=0.011). No differences were found with regard to death, rehospitalization, myocardial infarction, stroke or transient ischemic attack, systemic embolism or mitral thrombosis. Similarly, no statistically significant difference was found for the presence of AF (p for interaction=0.823).

Bleeding events were classified as 1 on the BARC criteria scale, and OAC with VKAs on discharge was predicted to generate three times more events. The bleeding rate in the VKA group was similar to the subgroup of patients with bioprosthetic valves included in the trials that compared NOACs to VKAs, although driven by minor bleeding events.16,19–22 This was probably related to the short follow-up period (within three months of surgery) in the study, since NOAC-related bleeding events tend to occur later in the course of treatment.23

The primary efficacy outcome in the ENAVLE trial (death, clinical thromboembolic event, or asymptomatic intracardiac thrombus) favored edoxaban (noninferior and superior) over warfarin, as did safety outcomes (bleeding events) and net clinical outcome, but without statistical difference (noninferior).17 These results were consistent with those in the present study, with some differences: our study showed NOACs to have comparable efficacy and better safety properties compared to VKAs, whereas ENAVLE showed better efficacy and similar safety in patients treated with NOACs; the lower rate of edoxaban use in the present study does not permit a direct comparison between the studies.

Analysis of the subpopulation of AF patients in the present study showed no differences in bleeding or stroke events in the NOAC group compared to VKA patients, although it is suggestive of fewer bleeding events in the former. In the RIVER trial,16 the primary outcome (net clinical outcome) was noninferior for rivaroxaban versus warfarin, and bleeding events also showed no differences between groups, although only 18.8% of patients were enrolled and randomized within three months of surgery. These findings were corroborated by the present subanalysis of AF patients.

Transition between different types of oral OAC may represent a period of increased risk of thromboembolism or bleeding, especially when switching from NOACs to VKAs.24,25 In the ROCKET-AF trial,22 there was an increase in thrombotic events after discontinuation of rivaroxaban and transition to warfarin at the end of the study. At the end of the ARISTOTLE trial26 (apixaban versus warfarin in AF patients), the blinded study drug was stopped, and open label warfarin was recommended. Patients who switched from apixaban to warfarin had more thrombotic and bleeding events. The higher risk of stroke and bleeding in these patients was probably associated with initiation of VKAs (comparable with the group on stable warfarin) rather than with stopping apixaban.24 Similarly, in the present study, switching OAC therapy on discharge in AF patients from NOACs to VKAs was associated with 12-fold higher odds of more bleeding (p=0.039).

While the current 2020 ACC/AHA guidelines provide no recommendation on OAC after MVR, the ESC/EATCS guidelines recommend thromboprophylaxis with VKAs in the first three months, as well as in patients undergoing mitral BVR.8,9 The clinical evidence for antithrombotic therapy after MVR is limited, based mainly on expert opinion, but the use of an annuloplasty ring and the possible occurrence of AF after surgery may justify short-term therapy.27 From a surgical standpoint, the present analysis showed that 64% of those in the NOAC group underwent MVR and 59.2% of the VKA group underwent mitral BVR (p=0.008), and the primary outcome showed no statistical difference between the type of surgery (p for interaction=0.954). NOACs thus appear to be a potential alternative to VKAs.

Although rate and rhythm management of rheumatic AF after valve intervention are suggested in the literature, understanding of the management of antithrombotic therapy after MVR or mitral BVR in RVD patients is limited.28 In this study, RVD was present in 26.7% patients, of whom 87.5% were discharged on VKAs (p=0.028). Nevertheless, bleeding (p for interaction=0.119) and stroke events (p for interaction=0.789) did not differ depending on the presence of RVD.

This was an observational study based on information obtained from medical records and by telephone consultation with patients, and is thus limited in the scope of collectable data. For example, lack of reporting of INR and time in therapeutic range is an acknowledged limitation. The presence of only minor bleeding events (BARC 1) may also limit the significance of the study's conclusions. Finally, the findings from this study, which recruited a small sample from a single center, may not be generalizable to other cardiac centers.

NOACs were associated with fewer events within three months of surgical MVR or mitral BVR, supporting their efficacy and safety compared to VKAs. In addition, recent data from large trials predict a promising role for NOACs used in the immediate postoperative period in these patients.

The authors have no conflicts of interest to declare.